Heads and Tails: An Analysis of Visual Signals in Cats, Felis catus

Citation and access

Bertrand L. Deputte, Estelle Jumelet, Caroline Gilbert and Emmanuelle Titeux (2021). Heads and Tails: An Analysis of Visual Signals in Cats, Felis catus. Animals, 11(9), 2752. DOI: 10.3390/ani11092752.

Original journal article · Legal open full text at PMC · Publisher PDF.

Published online 21 September 2021, in the September 2021 issue. Publisher PDF header, extracted publisher page and Crossref agree; a search-index rendering displayed a one-day-earlier date and is not used for the citation. The observations themselves were made in 2006, not 2021.

Coverage: Lumi’s AI-assisted synthesis of the accessible main text: Introduction, Methods, Results, Discussion, limitations and figure captions. Figure pixels/cell values, raw data and supplements were not independently audited. The publisher PDF header and abstract were additionally inspected. Open-access CC BY article; only an original summary is reproduced here.

Corrections and code: No formal correction identified in the inspected source material, targeted search or Crossref relation/update fields (empty/absent). The publisher version-notes page could not be directly retrieved, so its complete history is unverified. No author-linked implementation identified in the article material or targeted search; no repository inspection or code execution claimed. Reporting discrepancies below remain unresolved rather than silently corrected.

Why this paper matters — Lumi

A raised tail is conspicuous, but it is not a complete sentence. This study asks us to read ears, tail and interaction partner together, rather than treating one body part as a universal dictionary.

It connects cat–cat and cat–human communication in the same setting. Its practical value is a more careful observational habit, not a promise that we can read a cat’s thoughts or infer permission to touch from one posture.

Question and methods — Lumi

Question: Which combinations of ear and tail positions accompany approaches and their outcomes, and do patterns differ when the recipient is another cat or a human?

Researchers observed a semi-free-ranging shelter colony in France. They identified 29 relatively visible cats within a larger colony and completed 100 hours of observation across winter and summer periods in 2006. Sampling began when an identified cat approached another identified cat; people were instructed to stand still when approached, and human-initiated approaches were excluded.

At the start of cat–cat interactions, the observer recorded both partners’ ears (erect, flattened, or down/backwards) and tails (up or not up). A configuration means this combination of visible features. Tail motion was not coded. Notes were handwritten; there was no video recording.

Methods report 254 cat–cat interactions and 104 cat–human approaches. Outcomes were assessed for 196 cat–cat interactions, not all 254. Positive/neutral outcomes included contact, rubbing, licking, close proximity or simply passing on; negative outcomes included aggression, avoidance or flight. Thus, “positive/neutral” is a broad behavioral category, not a direct measure of happiness. Frequency-based statistical tests examined associations between configurations and outcomes.

Findings — Lumi

1. In most observed cat–cat encounters, both cats held their tails down. A tail-up display was therefore not necessary for every non-negative encounter.
2. Ear configurations were associated with the subsequent cat–cat outcome. Both partners having erect ears was associated with more positive/neutral outcomes; non-erect ears, especially in the receiver, were associated with more negative outcomes.
3. Tail-up approaches were more frequent toward humans than toward other cats in this sample. That is a context-specific difference, not a universal emotional translation.

Lumi’s interpretation: Observe the recipient as well as the approaching cat. A relaxed-looking initiator does not tell us how the receiver will respond. The study supports looking at combinations and context; it does not experimentally establish what cats perceive, prove that ear posture causes an outcome, or show that cats mentally classify humans as mothers.

Limitations and inference boundaries — Lumi

• Observation, not manipulation: Postures and outcomes covary. No controlled change to ear/tail displays established which feature the receiver actually used. “Predictor” here does not mean a validated out-of-sample emotion classifier.
• Repeated encounters: Interactions are not independent animals. The reported frequency analyses do not describe an individual/dyad random-effects model. A dyad means a pair of animals. Repeated cats and pairs, visibility-based selection and one shelter limit generalization.
• Incomplete recording: No video, no tail-motion analysis, only some outcomes classified, and sparse configurations pooled. The study is not a complete feline body-language atlas.
• Reporting discrepancies: The abstract says 10 cat–human interactions while Methods, Results and Figure 7 say 104. Methods and earlier Results say 254 cat–cat interactions, but section 3.3 and Figure 7 say 257. Section 3.3 describes 97.8% as tail-up/erect-ear approaches while combining codes 1100 and 1001; by the stated coding, 1001 includes backwards ears. Consequently, this summary does not treat that percentage as a verified estimate of erect-ear approaches or reconstruct a precise effect from inconsistent denominators.
• Caregiver hypothesis: The Discussion proposes a link to kitten–mother greeting behavior. This is an interpretation, not a tested finding about how cats conceptualize people.
• No automatic touch permission: An approach or tail-up posture does not establish that a particular cat welcomes handling. Welfare decisions require the whole situation and the animal’s ongoing responses.

Reading route — Lumi

Start with Methods 2.2: understand what counted as an approach and outcome. Then read Results 3.1–3.3, keeping cat–cat outcomes separate from cat–human approach frequencies. Finish with Discussion 4.2 and compare the numerical inconsistencies above before adopting a “tail-up means friendly” rule.

Guided reading — Lumi

This lesson follows the open main text; it does not claim an independent audit of figure images or raw observations.

1. Define the unit before interpreting the pattern

• What to look for: Twenty-nine identified cats, repeated interactions and 100 observation hours; 196 cat–cat outcomes assessed.
• What it means: An interaction count is not an animal count. The same cats and pairs can contribute repeatedly, and the outcome subset differs from the full sample.
• Why it matters: Numerous encounters can describe a colony well without supplying equally numerous independent examples of all domestic cats.

2. Read a configuration, not one isolated feature

• What to look for: Three ear categories, two tail categories and records for both cat partners at the start of an encounter.
• What it means: The comparison concerns combinations. The recipient’s posture supplies information the initiator’s tail alone cannot provide.
• Why it matters: A single-feature rule can miss conflicting cues. This study motivates a broader observation strategy, but it does not tell us precisely which cues a cat consciously attends to.

3. Separate the two comparisons

• What to look for: Cat–cat posture/outcome associations in section 3.2 versus tail positions toward cats or humans in section 3.3.
• What it means: One comparison concerns what follows an encounter; the other concerns how approaches differ by partner. They do not measure the same outcome.
• Why it matters: More tail-up approaches to people cannot, by itself, prove a stronger attachment bond, a maternal representation or willingness to be stroked.

4. Check the labels before trusting the headline number

• What to look for: The abstract’s 10 versus the main text’s 104 human approaches; 254 versus 257 cat–cat encounters; the mixed ear codes attached to 97.8%.
• What it means: The publication has unresolved reporting inconsistencies. The broad reported direction can be summarized while exact denominators and the ear-specific percentage remain uncertain.
• Why it matters: An honest reading preserves uncertainty rather than silently repairing the source. Better follow-up work could use video, independent coding and models accounting for repeated cats/pairs; these are suggestions, not procedures performed here.

Two reflection questions — Lumi

1. If an approaching cat has its tail up but the receiving cat’s ears move backwards, what would a tail-only interpretation miss?
2. How could a follow-up study distinguish a receiver’s response to ear position from its response to distance, prior relationship or the rest of the approaching cat’s posture?